Using Text Mining to Combine Unstructured Data with Structured Data


Published on

Combining customer "verbatims" with structured data

  • Be the first to comment

  • Be the first to like this

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide

Using Text Mining to Combine Unstructured Data with Structured Data

  1. 1. Management Consulting & Strategic CommunicationsBuild More Power into Customer Experience Efforts– UseText Mining of Unstructured Data to Complement Structured DataCapturing data from unstructured sources such as open-endedsurveys, phone and email support, and social media networks like But how can you combineTwitter and Facebook provides valuable opportunities for financial this unstructured data withservices organizations. This “voice” data offers direct, candid traditional structured data tofeedback from customers and prospects. gain a more complete picture of the customer experience?Leveraging all your customer data for a complete pictureCompanies have access to a wealth of structured data such as customer profiles, transaction histories,and surveys. Data mining and modeling of this data can yield actionable insights for issues such asattrition, customer service, and risk management. Adding unstructured text data, you can incorporatetext mining into the analysis, enabling you to extract complex concepts and analyze sentiment, toidentify service and operational issues, as well as emerging trends of concern, cross-sell opportunities,and more.Structured and unstructured data can each be used to enhance analysis of the other, or be combinedinto a common model:• Flags representing complex concepts and sentiments found in unstructured text can be added into a model with structured data to created a combined analysis.• Conversely, modeling of structured data can suggest a data subset that would benefit from text mining.As an example, modeling of structured data could identify that attrition of checking account customersis most significant for single males ages 18-25 who have less than 3 accounts. You could then use textmining on unstructured comment sources to uncover checking issues for that subset of customers, andgain better insight into their pain points. For instance, including in your analysis customer commentssuch as “dissatisfaction with new multiple ID requirements” and “difficulties balancing accounts usingonline banking” could provide valuable missing pieces to the attrition problem you’re trying to solve. 1
  2. 2. Management Consulting & Strategic CommunicationsCombining sources to maximize insights from social mediaWhile social media feedback such as tweets and Facebook posts are a key source for customerexperience insights, monitoring all of it can be overwhelming. Combined modeling of thisunstructured data with social media metadata –such as user profiles, location data, followers,demographic details, and cookies– can help you target the most important comments. Thisstrategy gives you a clearer picture of who’s saying what, and what it means, so you can respondmore effectively.Mining and analyzing structured and unstructureddata together can create a powerful lens into thecustomer experience, enabling you to make truly Let’s talk...informed decisions that will lead to better businessoutcomes. Toll-free: 1.877.676.3743 Website: Email:© 2011 Beyond the Arc, Inc. . 2600 Tenth Street, Suite 616 Berkeley, CA 94710 2